The aim of this study is to quantify the relative contributions of two muscle energy consumption processes (the detachment of cross-bridges and calcium-pumping) incorporated in a recently developed muscle load sharing cost function, namely the energy-based criterion, by using in vivo measured glenohumeral-joint reaction forces (GH-JRFs). Motion data and in vivo GH-JRFs were recorded for four patients carrying an instrumented shoulder implant while performing abduction and forward flexion motions up to their maximum possible arm elevations. Motion data were used as the input to the delft shoulder and elbow model for the estimation of GH-JRFs. The widely used stress as well as the energy-based cost functions were adopted as the load sharing criteria. For the energy-based criterion, simulations were run for a wide range of different weight parameters (determining the relative contribution of the two energy processes) in the neighborhood of the previously assumed parameters for each subject and motion. The model-predicted and in vivo-measured GH-JRFs were compared for all model simulations. Application of the energy-based criterion with new identified parameters resulted in significant (two-tailed p < 0.05, post-hoc power ∼ 0.3) improvement (on average ∼20%) of the model-predicted GH-JRFs at the maximal arm elevation compared to when using either the stress or the pre-assumed form of the energy-based criterion. About 25% of the total energy consumption was calculated for the calcium-pumping process at maximal muscle activation level when using the new parameters. This value was comparable to the corresponding ones reported in the previous literature. The identified parameters are recommended to be used instead of their predecessors. © 2013 World Scientific Publishing Company.